Determine the units of the horizontal axis. It may show the date and time of each run, but more commonly it gives the number of the control run.
Determine the units of the vertical axis. They should be the same as for the data itself; i.e., if the data is in meters, the numbers on the vertical axis should represent meters. You will notice that the numbers are probably not nice multiples of 2, 5, 10 or 100. Instead, each marked value is one standard deviation away from the previous one, and the standard deviation could be any number, like 16.4 or 0.235, and is more likely to be "messy" than not.
Determine where the mean is. Usually a key on the graph will tell you which line indicates the mean. Often, different styles of dashed lines indicate the mean, the first standard deviation, the second standard deviation and, if it is on the graph, the third standard deviation.
Count how many of the data points lie within the first standard deviation, then within the second (not including those in the first), then the third (not including those in the first or second). This will give you a rough estimate of where most of the data lies.